Re: 1T and 500+500 mirror, but different speeds

On 11/07/21 04:12, David T-G wrote:

Wol, et al --

...and then Wols Lists said...
% 

% 
% Mirroring as you plan here is okay.

That's good.  Of course, that then makes me scratch my head a bit since
it's still paired with other and needs to not blow up ...  I'll have to
back and read more to see why mirror is safe.

Mirroring is safe because, whatever happens to the mirror, you only need
one disk to recover. So if the raid falls apart your data is not in danger.

% 
...
% > and also two 500G drives
% > 
% >   jpo:~ # parted /dev/sdc print
% >   Model: ATA WDC WD5000AAVS-0 (scsi)
...
% >   jpo:~ # smartctl -i /dev/sdc
...
% >   Model Family:     Western Digital Caviar Green
...
% > that I plan to stripe together to use to protect the first.  The Caviars
% > are listed as variable speed, but in practice they apparently are just
% > 5400 rpm, so I'd like to take advantage of striping to make them as fast
% > as possible. 
% 
% Caviar Green ??? Caviars should be okay, but the "green" moniker makes

Yeah, I'm not too happy about that ...  Supposedly they will vary speed
based on demand, so they can deliver 7200rpm performance, but apparently
they always stay at 5400rpm and so they're just basic drives.  Meh.


% me nervous. Check SCT/ERC, but striping/mirroring will be fine.

More of that confusion stuff :-) but happy.


% 
% > This isn't the black magic ;-) of RAID10-on-two-drives,
% > so I don't have to think of one as "front" and one as "back", but do I
% > want more than one partition on each to stripe 4 or 6 slices to avoid
% > hammering on one, or do I just go with each device as a whole and let
% > mdadm handle the magic for me?
% > 
...
% Okay, the simple approach.
% 
% Create a single-device mirror using the 1TB and the special device
% "missing". Copy everything across and make sure it's all okay. (I'm
% assuming you can safely wipe this drive as it has nothing on it you wish
% to keep.)

Correct.


% 
% Create a striped device using the two 500GB drives.

That's the fun part ...  I'm thinkin' I'm just going to have to do sector
math to predict how large the new dev can be so that I can back into a
proper partition size on the 1T so that both halves can match.

Do you need the entire space? Can you create a 900GB mirror on the 1TB?
You can always grow the mirror and the filesystem later.

% 
% Plan to replace all the disks with something like Seagate Ironwolves or
% Toshiba N300s in the near future :-)

Actually, I have four Seagate 4T drives

  diskfarm:~ # parted /dev/sdd print
  Model: ATA ST4000DM000-1F21 (scsi)
  Disk /dev/sdd: 4001GB
  Sector size (logical/physical): 512B/4096B
  Partition Table: gpt
  Disk Flags: 

  Number  Start   End     Size    File system  Name                             Flags
   1      1049kB  4001GB  4001GB  xfs          ata-ST4000DM000-1F2168_W300EYNA  raid
   2      4001GB  4001GB  134MB   reiserfs     wwn-0x5000c50069a8d76f

  diskfarm:~ # smartctl -i /dev/sdd
  smartctl 7.0 2019-05-21 r4917 [x86_64-linux-5.3.18-lp152.63-default] (SUSE RPM)
  Copyright (C) 2002-18, Bruce Allen, Christian Franke, www.smartmontools.org

  === START OF INFORMATION SECTION ===
  Model Family:     Seagate Desktop HDD.15
  Device Model:     ST4000DM000-1F2168
  Serial Number:    W300EYNA
  LU WWN Device Id: 5 000c50 069a8d76f
  Firmware Version: CC52
  User Capacity:    4,000,787,030,016 bytes [4.00 TB]
  Sector Sizes:     512 bytes logical, 4096 bytes physical
  Rotation Rate:    5900 rpm
  Form Factor:      3.5 inches
  Device is:        In smartctl database [for details use: -P show]
  ATA Version is:   ATA8-ACS T13/1699-D revision 4
  SATA Version is:  SATA 3.1, 6.0 Gb/s (current: 6.0 Gb/s)
  Local Time is:    Sat Jul 10 21:38:12 2021 UTC
  SMART support is: Available - device has SMART capability.
  SMART support is: Enabled

in another machine that I plan to upgrade to larger disks and hand down
to this one.  They're older, and they don't

  diskfarm:~ # smartctl -l scterc /dev/sdd
  smartctl 7.0 2019-05-21 r4917 [x86_64-linux-5.3.18-lp152.63-default] (SUSE RPM)
  Copyright (C) 2002-18, Bruce Allen, Christian Franke, www.smartmontools.org

  SCT Error Recovery Control command not supported

have SMART error handling, but at least they aren't SMR :-)

They're no better! "SCT Error Recovery Control command not supported".
This is the red flag. This is the same problem as with the Barracudas.

% 
% https://raid.wiki.kernel.org/index.php/Linux_Raid
% 
% Read especially the page on timeout mismatch as this DOES apply to your
% Barracuda !!!

I'll go back again :-)

And it applies to your 4TBs as well. I wouldn't worry too much - I'm
running a mirror on two 3TB Barracudas (I wouldn't have bought them if
I'd realised, but I was a raid newbie back then ...)

Cheers,
Wol